Radiographic image acquiring device, radiographic image acquiring system, and radiographic image acquisition method
A image acquiring device includes a camera configured to scan radiation passing through a target object in one direction and acquire an X-ray image, a scintillator configured to convert the X-rays into light, and a control device configured to input the X-ray image to a trained model constructed through machine learning in advance and execute a noise removal process. The camera includes a scan camera in which pixel lines each having M pixels arranged in one direction are configured to be arranged in N columns in a direction orthogonal to one direction and which is configured to output a detection signal for each of the pixels, and a readout circuit configured to output the X-ray image by adding the detection signals output from at least two pixels for each of the pixel lines of N columns in the scan camera and sequentially outputting the added N detection signals.
1. A radiographic image acquiring device comprising:
an imaging device configured to scan radiation passing through a target object in one direction and capture an image thereof to acquire a radiographic image;
a scintillator configured to be provided on the imaging device to convert the radiation into light; and
an image processing module configured to calculate an average energy related to the radiation passing through the target object, evaluate the spread of a noise value using the average energy, input the radiographic image to a trained model constructed through machine training in advance using image data and execute a noise removal process of removing noise from the radiographic image,
wherein the imaging device includes
a detection element in which pixel lines each having M (M is an integer equal to or greater than 2) pixels arranged in the one direction are configured to be arranged in N columns (N is an integer equal to or greater than 2) in a direction orthogonal to the one direction and which is configured to output a detection signal related to the light for each of the pixels, and
a readout circuit configured to output the radiographic image by adding the detection signals output from at least two of the M pixels for each of the pixel lines of N columns in the detection element and sequentially outputting the added N detection signals, and
wherein the image processing module includes
at least one processor configured to derive an evaluation value obtained by evaluating a spread of a noise value from a pixel value of each pixel of the radiographic image on the basis of relational data indicating a relationship between the pixel value and the evaluation value and generate a noise map which is data obtained by associating the derived evaluation value with each pixel of the radiographic image, and
input the radiographic image and the noise map to the trained model and execute the noise removal process of removing noise from the radiographic image.
2. The radiographic image acquiring device according to claim 1 , wherein the trained model is constructed through machine training using image data obtained by adding a noise value along a normal distribution to a radiographic image of a predetermined structure as training data.
3. The radiographic image acquiring device according to claim 1 , wherein the image processing module includes
at least one processor configured to accept an input of condition information indicating either conditions of a source of radiation or imaging conditions when the radiation is radiated to capture an image of a target object,
calculate average energy related to the radiation passing through the target object on the basis of the condition information, and
narrow down trained models used for the noise removal process from a plurality of trained models constructed through machine training in advance using image data on the basis of the average energy.
4. The radiographic image acquiring device according to claim 1 , wherein the image processing module includes
at least one processor configured to specify image characteristics of the radiographic image acquired by the imaging device with a jig as a target,
select a trained model from a plurality of trained models constructed through machine training in advance using image data on the basis of the image characteristics, and
execute the noise removal process using the selected trained model.
5. A radiographic image acquiring system comprising:
the radiographic image acquiring device according to claim 1 ;
a source configured to radiate radiation to the target object; and
a transport device configured to transport the target object to the imaging device in the one direction.
6. A radiographic image acquisition method comprising:
scanning scintillation light corresponding to radiation passing through a target object in one direction and capturing an image thereof to acquire a radiographic image;
calculating an average energy related to the radiation passing through the target object and evaluating the spread of a noise value using the average energy; and
inputting the radiographic image to a trained model constructed through machine training in advance using image data and executing a noise removal process of removing noise from the radiographic image,
wherein the acquiring of the radiographic image includes using a detection element in which pixel lines each having M (M is an integer equal to or greater than 2) pixels arranged in the one direction are configured to be arranged in N columns (N is an integer equal to or greater than 2) in a direction orthogonal to the one direction and which is configured to output a detection signal related to the scintillation light for each of the pixels, to output the radiographic image by adding the detection signals output from at least two of the M pixels for each of the pixel lines of N columns in the detection element and sequentially output the added N detection signals, and
wherein the executing includes deriving an evaluation value obtained by evaluating a spread of a noise value from a pixel value of each pixel of the radiographic image on the basis of relational data indicating a relationship between the pixel value and the evaluation value, generating a noise map which is data obtained by associating the derived evaluation value with each pixel of the radiographic image, inputting the radiographic image and the noise map to the trained model, and executing the noise removal process of removing noise from the radiographic image.
7. The radiographic image acquisition method according to claim 6 , wherein the trained model is constructed through machine training using image data obtained by adding a noise value along a normal distribution to a radiographic image of a predetermined structure as training data.
8. The radiographic image acquisition method according to claim 6 , wherein the executing includes accepting an input of condition information indicating either conditions of a source of radiation or imaging conditions when the radiation is radiated to capture an image of a target object, calculating average energy related to the radiation passing through the target object on the basis of the condition information, and narrowing down trained models used for the noise removal process from a plurality of trained models constructed through machine training in advance using image data on the basis of the average energy.
9. The radiographic image acquisition method according to claim 6 , wherein the executing includes specifying image characteristics of the radiographic image acquired with a jig as a target, selecting a trained model from a plurality of trained models constructed through machine training in advance using image data on the basis of the image characteristics, and executing the noise removal process using the selected trained model.
10. The radiographic image acquisition method according to claim 6 , further comprising:
radiating radiation to the target object; and
transporting the target object to the detection element in the one direction.